Re: Roadmap to 3.0
On Wed, Sep 25, 2002 at 07:45:36AM +0200, Eddie Lania wrote: Loadable RPC implementations yes o o o Does this mean that it will support KIXTART as well? I've yet to see a good definition of what KIXTART needs, so I can't say if we will support it. Andrew Bartlett
Re: strupper() on username and domain in cli_session_setup_nt1()?
On Mon, Sep 23, 2002 at 08:30:28AM -0500, Gerald Carter wrote: On Mon, 23 Sep 2002, Tim Potter wrote: I'm just wondering whether it is strictly necessary to uppercase the username and domain name when performing a session setup. Applying the following patch should not break anything and would make smbclient be able to test for case sensitivity bugs in the remote server. The behaviour of uppercasing everything seems to be peculiar to Windows 9x. According to some testing done by the HP guys and girls, NT and 2000 send across the case of the username as it was created in the SAM, whereas XP sends the case the user logged in with. The patch looks fine to me Tim. Makes sense to me - particularly when the remote server is a samba box with an InterCase username. Andrew Bartlett
Re: approaching release of 3.0alpha20
On Mon, Sep 23, 2002 at 10:01:53AM -0500, Gerald Carter wrote: Everyone, I would like to do another alpha snapshot release of the 3.0 code base later this week. Does anyone know of any code that is too unstable for a release (seg faults, etc...)? Apparently we ignore non-primary groups (which, given recent changes I can certainly beleive), but I've not had a chance to chase it up. Andrew Bartlett
Re: Samba 3.0a19 breaks winbind helpers?
On Tue, Sep 10, 2002 at 11:37:38AM +0200, Chemolli Francesco (USI) wrote: This program is intended to be invoked directly by external programs, so they can pass in the string directly into the right argv buffer. Shell users can quote. What about a also supporting a stream oriented NTLM mode? I don't see the need - most applications doing this so frequently that they need a stream mode are doing NTLMSSP anyway. Less interfaces again... I think a stream interface is needed for performance reasons. Why? We have NTLMSSP as a stream - and if I don't currently see any application for raw NTLM outside the MSCHAPv2 example - which probalby would not benifit anyway... I think the performance issues here are overrated. Andrew Bartlett
Re: initial gencache implementation
On Fri, Sep 06, 2002 at 10:17:19AM +1000, Tim Potter wrote: On Thu, Sep 05, 2002 at 12:15:36PM +0200, Rafal Szczesniak wrote: This is first implementation of caching mechanism. It includes both lib/gencache.c code and utils/net_cache.c as command-line control/testing tool. comments are welcome Rafal, that looks pretty good. Since you ask, I do have a few comments. (-: You assume that any cached data will be in null terminated string format which is not always the case. I understand this is a design property - it's up to the caller to mess with structs etc. This keeps the cache simple, and allows for human inspection with 'net cache' etc. This is just my personal opinion but I would prefer for gencache_set to crash if you pass it a NULL pointer for the key or value parameter. Returning false in this case only hides the error until further along in the execution path by which time it will be more difficult to track down the original error. Agreed. Andrew Bartlett
Re: samba internals and application development advice
On Wed, Aug 21, 2002 at 06:31:57PM -0400, Sean Finney wrote: hi all-- i've been ghosting this list for a few months now, hoping that i might read an answer to my question without having to step up and ask, but now i'm guessing that this won't be the case, so here goes... i'd like to write an application in C that does something along the lines of crawling a local area network, finding smb sharing computers and enumerating shares. See smbtree, in most recent Samba versions. Andrew Bartlett
Re: Thanks for fixing oplock.c for Linux 2.0 in 2_2 CVS
On Wed, May 29, 2002 at 04:43:05PM -0700, Jeremy Allison wrote: On Thu, May 30, 2002 at 08:10:22AM +1000, Andrew Bartlett wrote: Isn't there a way we can 'idle' the connection by tearing down the protocol? Actually issuing a 'you are idle, shutting down' to the client? Nope - would require a client change I'm afraid. There's nothing in the protocol to allow this. What about at the NetBIOS level? I'm wondering how somthing like NetBEUI (being just raw netbios on ethernet effectivly) closes connections. And are you saying that Win2k will never 'idle' a client connection? I'm sure I've seen smbfs being 'idled' by NT before... Andrew Bartlett
Re: [PATCH] store SID's in SAM_ACCOUNT
On Tue, May 28, 2002 at 02:07:52PM +0200, Simo Sorce wrote: Hi Stefan. As you may have seen I have already changed the pdb_interface to search by SID and I'm really i favour to use SIDs inside SAM_ACCOUNT instead of RIDS, but I think this patch does not address the problem the right way. What we should do is store the SID in the backends, not convert it at run time. Given time, yes. But (for example) we really cannot change the LDAP schema at this stage (we must certainly still support the current schema) and other backends might chose to store the RID/SID in a way that best suits their technology. I think we may use part of your code inside pdbedit to have a tool to upgrade from previous backends that store by RID into the new ones. I don't see the need for a forced 'upgrade'. New backends can use this - but with the majoir passdb backends (ldap, smbpasswd) still being RID based into the future, we are going to still need this. I'm working on tdbsam2 that will store by SID and have also some other interesting things (privileges and such) I have discussed with JFM the last samba experience conference. BTW, pdbedit already handles 'upgrades', just import from one format and export to the other. This code will ensure that it all 'just works' What others do think? I think this patch is fine. We have to start somewhere - and while you might *like* to rework the while passdb in one hit, I don't see any reason why an incremental patch cannot be applied in the meantime. On the patch itself, my only comment is that when implementing the compatibilty functions, make them call the 'real' funcitons (see pdb_set_plaintext_password()) rather than modifiying the struct directly. Also I see you still have pdb_set_user_rid(), which just calls the fallback. If this is really a complete patch, then these functions can go, and other coders will notice it - and hopfully see what the new interface is. Either that, or don't rename the fn at all. Finally, re the SID initialisation: Make a new function that returns the global sam sid. Make it 'auto-initailise', if its not been run before, then it should figure it out, and store it in a static. Andrew Bartlett
Re: Proposed code cleanup
On Fri, May 24, 2002 at 11:27:39AM +1000, Tim Potter wrote: On Fri, May 24, 2002 at 11:17:59AM +1000, Andrew Bartlett wrote: Anyway - that looks like a good patch. I've been dying to get rid of all those silly make_nmb_name() calls everywhere but was not brave enough. I'll proabably add a cli_pipe_connection() function (feel free to suggest a better name) that just calls cli_full_connection, but with IPC$ implicit and instead specifying the pipe to setup. How does that sound? Sounds good to me. One point - is there any reason to pass around the password and the length of the password to cli_full_connection()? Not really. Just habit - it is always a null terminated ASCII (or utf8 actually) password. (and everyobdy is just doing strlen for that param anyway...) I'll kill it. Andrew Bartlett
Re: winbindd uid and gid range assumptions
On Tue, May 14, 2002 at 08:20:03AM -0400, Mike Gerdts wrote: On Mon, 2002-05-13 at 18:42, Andrew Bartlett wrote: Moving over the socket is a very expencive operation, particularly compared to a simple if statement. Also, where we know that a uid is local, we need to check with code that winbind isn't linked to - the passdb backend. So in a situation where you have UIDs interspersed between NIS and domain users, it may be cheaper to check to see if it is local first followed by winbind. At least this may be better in my situation, as each of my NFS/Samba servers is already an NIS slave. Even though a UID lookup may have to talk to nscd and/or ypserv, it is still on the same machine, thus avoiding network delays. Perhaps this would be a place where the plug-in architecture could be useful as well. Checks could all be relegated to idmap_ops-islocal(). The default op could be to check the winbind id range. Others that are willing to or need to pay the price of a socket operation will have the option of doing so. Presumably islocal() would not just be a straight BOOL operation. I could imagine it replying True, False, LocalFirst, or DomainFirst. Sounds like an interesting idea. But yes, we need to deal with things like getting the uid from the SFU LDAP schema, so this may well change in the future. Do you have any relative time frame or rough release number that you are shooting for? No, its just an idea in my HEAD until I get a chance to look at it, or sombody codes me up a patch :-). Do you see a plug-in that merges the functionality of the existing idmap to the architecture present in the VFS, or should I start barking up a different tree? That seems to be the current plugin arch, but we may move across to a more centralised scheme at some stage. Andrew Bartlett
Re: [Samba] Impending Removal of --with-ssl
On Sun, May 05, 2002 at 10:06:53AM -0400, Nathan Lutchansky wrote: On Sun, May 05, 2002 at 02:50:13AM -0700, [EMAIL PROTECTED] wrote: On Sat, May 04, 2002 at 11:22:41PM -0400, Nathan Lutchansky wrote: 1) Can we assume that Microsoft will never include SSL functionality in their Windows clients? Does MS have some other method of providing transport security instead? If the answers are yes and yes, then I'd say it is safe to remove. Otherwise it might feel silly to add SSL back when some XP service pack adds SSL functionality later on. yes on both counts. Message authenticaion and encryption are available in the CIFS protocol, and are detailed in the SNIA Technical Reference (not to be confused with the MS Technical Reference) Oh. Well, that sounds like the way to go in the future. I hope it is not as ugly to implement as SSL. 2) I'd started a project to authenticate users SMB clients based on client SSL certificates. If --with-ssl is removed, SSL authentication can still be done with wrappers and LIBSMB_PROG, but the server wrapper would somehow need to pass authentication information to Samba. The easiest way is to setreuid to the target user before execing smbd, but can smbd handle this? What happens if smbd is started (without -D) as some user other than root? -Nathan Samba expects this, and allows become_user() calls to 'fail' but still requires passwords as before. You could write a new authentication module that implments your requirements quite trivially. (And use environment variables or the like to pass the state info along). OK, I'll look into this when I have time to get back to that project. Thanks for the hint. While samba will 'cope' with non-root setups, this really only works in testing environments, where that same user owns the critical files. As such I would suggest you make your SSL wrapper leave smbd as root, and make a cusome authenticaion module figure it out from there. See samba's rhosts support module for a trivila example of what you want to do. (It only still exists becouse its a good example, not becouse anybody should use it...) Andrew Bartlett
Re: wbinfo -t and Win2K DCs ...
On Fri, May 03, 2002 at 09:38:33AM +0930, Richard Sharpe wrote: Hi, I have run into a spot of bother with wbinfo and Win2K. We run Samba as a member server from a Win2K DC. We join the DC correctly, but wbinfo -t does not work. Then you have not joined the domain corrrectly... This seems to be because the WINS server running on the Win2K DC will not answer queries looking for DOMAIN#1c. However, I know we have joined because wbinfo -u works. Is this in Samba 2.2 or 3.0? wbinfo -u works without a correct domain join in 2.2 - becouse the operations are done anonymously. Has anyone seen this? Is there a solution on the DC, or do I have to hack winbindd to be able to look up the DC correctly? I think you have not joined the domain correctly - it can't be wins if wbinfo -u is working. Andrew Bartlett
Re: copying 2Gb files
On Fri, May 03, 2002 at 09:54:44AM +0100, Mike Pigott wrote: Hello, We are running Samba 2.0.7 for IRIX (v6.5.11m) and are receiving the error cannot copy file * parameter is incorrect when attempting to copy files 2Gb. OS is Win98. Can you confirm if this is a known bug, or shall I triple check the code of the developer who is attempting to copy this data with his utility? Large file support really only grew to maturity in the 2.2 series - I would suggest an upgrade. Andrew Bartlett
Impending Removal of --with-ssl
This message is a warning: --with-ssl will die. Ok, thats enough with the dramatics, but the general consensus amoungst the samba team is that --with-ssl really isn't a particulary smart idea, and it is better implmented by external tools. So what is --with-ssl exactly? And why kill it? --with-ssl allows Samba to tunnel SMB inside an SSL connection. Unfortunetly there are only 2 clients: smbclient and sharity. Windows clients simply don't know how to use SSL. So why kill it? It might be useful to sombody? While some small minority of users might find it handy, it confuses many more, including a supprising number of our distributors. Users actually using this functionality will find that they can achive almost the same effect by creative use of 'stunnel' both as an inetd wrapper as as a 'LIBSMB_PROG' program. Finally, it is intrusive and ugly, with large #ifdef sections in what should be simple code. If sombody can come up with both reasons to keep this code, and time to maintain it, then I would like to hear it. Andrew Bartlett
Re: overriding dyn_CONFIGFILE in pdbedit with command line parameter
On Mon, Apr 29, 2002 at 05:00:32PM -0400, Bradley W. Langhorst wrote: I decided to modify pdbedit to handle another command line parameter -c /pathto/smb.conf I've got that working - but I'm not sure it's safe Is it reasonable to expect the popt stuff to give me back null terminated string? Can you resend as a diff -u? As to null termination, I think that is safe. Andrew Bartlett
Re: Re: User Manager for Domains / RAS callback problem
On Fri, Apr 26, 2002 at 10:35:41AM +0200, Jean Francois Micouleau wrote: On Thu, 25 Apr 2002, Alex Keahan wrote: For the time being, I am going to implement a temporary solution, but I would really like to see this fixed in the 2.2 branch. it won't be fixed in 2.2, it's way too complex. Can the callback info be easily added to the mix without pulling in the whole nine yards? If yes, what are the missing RPC calls that I should implement, and where do you think the extra info should go -- the smbpasswd file? What is the approach taken by the 3.0 / TNG branches? that's not only implementing some missing rpc calls, that's also changing the /etc/smbpasswd file format to store the callback number. This is where LDAP shines. I think we already store the callback number but there may be other RPCs required. I like the way we can easily extend the LDAP stuff - much better than smbpasswd/tdbsam in this respect. Andrew Bartlett